Quantum stereographic projection and the homographic oscillator
T. Hakiog˘luPhysics Department, Bilkent University, Ankara 06533 Turkey M. Arik
Physics Department, Bog˘azic¸i University, Istanbul 80815 Turkey ~Received 12 October 1995!
The quantum deformation created by the stereographic mapping from S2toC is studied. It is shown that the resulting algebra is locally isomorphic to su~2! and is an unconventional deformation of which the undeformed limit is a contraction onto the harmonic oscillator algebra. The deformation parameter is given naturally by the central invariant of the embedding su~2!. The deformed algebra is identified as a member of a larger class of quartic q oscillators. We next study the deformations in the corresponding Jordan-Schwinger representation of two independent deformed oscillators and solve for the deforming transformation. The invertibility of this transformation guarantees an implicit coproduct law which is also discussed. Finally we discuss the analogy between Poincare´’s geometric interpretation of the quantum Stokes parameters of polarization and the stereo-graphic projection as an important physical application of the latter.@S1050-2947~96!00106-0#
PACS number~s!: 03.65.Fd, 02.20.Sv, 02.20.Qs
I. INTRODUCTION
Since the discovery of the first deformed quantum algebra by Biedenharn and Macfarlane @1# a tremendous effort has been made to find physical realizations of these algebras. Much earlier, inspired by an operator representation for
q-deformed dual resonance models@2#, Baker, Coon, and Yu
formulated the simplest q algebra which produced a suitably bounded spectrum determined by the parameter of deforma-tion q. Apart from their profound mathematical significance mainly related to the solution of the quantum Yang-Baxter equation @3#, and other problems @4#, physical applications can be found in the deformed Jaynes-Cummings model@5#, the ubiquitous quantum phase problem @6#, the relativistic
q oscillator @7#, and recently, in reproducing deformed
nuclear energy levels @8#, the Morse oscillator @9#, and the Kepler problem@10#.
In particular, several possible realizations of deformed Lie algebras can be constructed by applying certain nonlinear invertible deforming transformations to the generators of the undeformed algebras @11#. In this context, some explicit cases have been examined by Curtright and Zachos @12#, among several previously studied examples.
The quantum deformation of a physical symmetry should be identified by a deformation parameter which must be uniquely determined by a set of observables. In this work, we give a particular example of that by examining the ste-reographic projection of the su~2! generators on an extended complex plane and show that the resulting deformation is described by a deformation parameter which is directly con-nected with the central invariant of the embedding su~2!. In Sec. II we define and derive the properties of the quantum stereographic projection. Section III is devoted to the prop-erties of the central invariant. In Sec. IV the deformation induced by the homographic oscillator on su~2! is studied using Jordan-Schwinger representation. The proof of exist-ence of the coproduct for the corresponding su~2! deforma-tion is presented in Sec. V. In the last secdeforma-tion, Sec. VI, we
discuss possible physical applications where quantized ste-reographic projection and the resulting homographic oscilla-tor algebra become relevant.
II. STEREOGRAPHIC PROJECTION
Stereographic projection~SP! is a mapping from the Rie-mann sphere S2 onto an extended complex plane C. At the
classical level, SP is defined by a mapping from S2in
spheri-cal J,u,fparametrization to that z,z*on the complex plane given by z52Jcotu/2 eif, with Su5 2v 11v2, Cu5v 221, v211, where v5
A
z*z 2J , ~1!where J is fixedu andf are real coordinates describing the polar and azimuthal angles, Su and Cu describe the sine and cosine of u, respectively, and z,z* are defined on the pro-jected plane as shown in Fig. 1. SP is an invertible mapping except for the ideal point at infinity. However, this does not violate the formal equivalence between the two representa-tions; since, as will be shown later, when Eq. ~1! is quan-tized, the ideal point at infinity is well represented by infinity in the discrete spectrum of the corresponding deformed alge-bra. We now proceed by defining an operator realization of Eq. ~1! via sine-cosine operators Cˆu and Sˆu as
Cˆu5Vˆ 221 Vˆ211, Sˆu5 2Vˆ 11Vˆ2, where Vˆ5
A
Zˆ†Zˆ, Zˆ5EˆfVˆ, ~2! 54and Vˆ and Eˆf are operator counterparts of v and eif, re-spectively. As in the case of angular decomposition of su~2!, here we deal with ideal unitary polar operators ~i.e.,
Cˆu21Sˆu251 and @Cˆu,Sˆu#50) whereas the azimuthal phase operator Eˆf has both a nonunitary as well as a unitary rep-resentation.
Throughout this work we assume that a quantum defor-mation is understood explicitly in the same sense as Refs.
@11,12#. To be more precise, providing invertible nonlinear
functionals of generators, the deformed algebraic structure and its representations are obtained by directly applying these nonlinear functionals to a particular representation of the undeformed mother algebra su~2! @11#. Depending on which level this substitution and following quantization take place, one naturally obtains different quantum deformations.
Using Eqs.~2! it can be found that
Zˆ†Zˆ511Cˆu 12Cˆu, ZˆZˆ †5Eˆ f 11Cˆu 12CˆuE ˆ f †. ~3!
In deriving Eq.~3! we did not assume the existence of any particular unitary representation for Eˆf~i.e., Eˆf†ÞEˆf21). The algebraic structure of the Zˆ,Zˆ†operators, however, as will be discussed later, is not influenced by any conflict between the unitary and nonunitary description of Eˆf. The algebra de-fined by Zˆ,Zˆ† can be found by using the well-known su~2! relation@13,14# @Cˆu,Eˆf#52 1 JE ˆ f ~4!
in Eq. ~3!, which is expressed by the generalized
commuta-tion relacommuta-tion
aˆaˆ†5paˆ †aˆ11
kaˆ†aˆ11~11@Eˆf,E ˆ f †#!, with aˆ51 aZˆ, aˆ†5 1 a*Zˆ†, ~5!
where aˆ(aˆ†) represent the normalized annihilation~creation! operators and the parameters q,k,a are given by
uau25u2J21u21, p5~2J11!/~2J21!,
k52~2J21!22. ~6!
Equation~5! is not an example of prototype deformed alge-bras and has not been studied in the literature. The commu-tation@Eˆf,Eˆf†# naturally arises in the derivation. It has been
recently shown that it is possible to find a manifestly unitary description of Eˆf without affecting any of its operator prop-erties@6#. Here, Eˆfis identical to the azimuthal phase opera-tor in the su~2! polar decomposition @6,15# and its unitary representation has the cyclic property
Eˆf5
(
m52 j j
u jm21
&^
j mu1bu j2 j&^
j ju, @Eˆf,Eˆf†#50~7!
in the finite dimensional Hilbert space spanned by the basis vectors u jm
&
. Here ubu51 and is otherwise undetermined, referring to an arbitrary reference phase. On the other hand, we must mention as a side remark that, if one adopts the nonunitary description of Eˆf @i.e. b50 in ~7!#, the first de-formed excitation energy of the algebra ~5! is influenced by the nonunitarity of Eˆf in such a way that it produces a scale transformation on the operators aˆ,aˆ†. However, its effect can always be absorbed by a further trivial renormalization of these operators. We will not elaborate on the other implica-tions of the nonunitary description of the Eˆf operator in this work.Equation~5! is actually in the class of generalized quartic oscillators@16# whose properties cannot be simply obtained by taking the square of any generalized commutator. Be-cause of the homographic dependence of aˆ†aˆ on aˆaˆ† we term the algebra in Eq.~5! a homographic oscillator ~HO!. In the limit J→` we observe the limits p→1 and k→0, there-fore HO contracts to the simple harmonic oscillator ~SHO!. This is reminiscent of the I˙no¨nu¨-Wigner contraction of su~2! onto the SHO @17#. The spectrum of HO can be solved ex-actly by considering a generalized Hermitian number opera-tor Nˆ such that @aˆ,Nˆ#5aˆ;@aˆ†,Nˆ #52aˆ†. For the most gen-eral solution, we have aˆ†aˆ5@Nˆ# and aˆaˆ†5@Nˆ11# where @Nˆ# is the principal number operator, and un
&
J are the basisvectors such that
@n11#5pk@n#11@n#11, ~8!
where
aˆun
&
J5@n#1/2un21&
J, aˆ†un&
J5@n11#1/2un11&
J,Nˆ un
&
J5nun&
J. ~9!Enforcing the ground state u0
&
J to be annihilated by aˆ, wehave@0#50. Using this ground state in ~8!, the whole spec-trum can be analytically iterated to yield
@n#5@@n##2@@n21##@@n## , @@n##5r1 n2r 2 n r12r2 ~r 1Þr2!, ~10!
where@0#5@@0##50 and @1#5@@1##51, with 1 r11 1 r2511p, 1 r1r25p2k. ~11!
From Eqs.~6! and ~11! we find r15r25q. The spectrum is thus given by the first derivative of@@n## with respect to q as FIG. 1. Geometric interpretation of stereographic projection.
@@n##5nqn21. Here we identify q5(2J21)/2J as the
de-formation parameter of the HO algebra.
It is known that the basic number @@n## arises in the so-lution of the generalized Fibonacci series@16#
@@n12##5a@@n11##1b@@n##,
where
a5r11r2, b52r1r2. ~12!
Further, it can be shown that Eq. ~12! defines a class of
generalized~i.e., r1Þr221) Biedenharn-Macfarlane~BM! ~or
Fibonacci! q oscillator. The commutation relation which yields ~12! can be found if two new operators bˆ,bˆ† are de-fined such that bˆ†bˆ5@@Nˆ##,bˆbˆ†5@@Nˆ11## as
bˆbˆ†5r2bˆ†bˆ1r1Nˆ. ~13!
In principle, Eqs.~10!, ~12!, and ~13! plausibly suggest that Eq.~5! can be effectively obtained from the generalized BM
q oscillator by a second deformation. Although a direct
transformation from one into the other has not been found, recently an attempt has been made to unify all quartic oscil-lators. In this scheme, ~5! and ~13! correspond to special cases such as r25r121, r25r12, or r25r1 ~see Ref. @18#!. This can be shown by applying the nonlinear transformation
bˆ5b˜ˆ ~b1g˜bˆ˜bˆ†!1/2, ~14! where@b˜ˆ,Nˆ #5b˜ˆ in Eq. ~13!, we get the form
Ab˜ˆ˜bˆ†˜ˆb˜bˆ†1Bb˜ˆ˜bˆ†˜bˆ†˜bˆ 1Cb˜ˆ†˜bˆb˜ˆ†b˜ˆ
1Db˜ˆ˜bˆ†1Eb˜ˆ†˜bˆ 1F50, ~15!
with the coefficients
A5g, B50 , C52qg,
D5b, E52qb, F521, ~16!
whereas the HO corresponds to the special case of the gen-eralized quartic oscillator in Eq.~15! with the coefficients
A50 , B5k, C50, D51 , E52q, F521.
~17!
Both Eqs.~16! and ~17! have the property AE5CD which is possessed by the quartic square root oscillator@16# as a spe-cial case of ~15!. In the limit J→` both homographic and Fibonacci oscillators contract to SHO.
III. CENTRAL INVARIANT
The HO algebra in ~5! is isomorphic to its underlying su~2! algebra. The range of values which the quantum num-ber n can take is limited by the total angular momentum J
~i.e., 0<n<2 J). This can be seen easily by causing the
diagonal operator Cˆu to act on the angular momentum
uJm
&
and, simultaneously, on the homographic oscillatorun
&
J bases. The action of the aˆ,aˆ† operators on the basis
vectors generates lower and upper bounds for its energy spectrum ~i.e., aˆu0
&
J5aˆ†u2J&
J50). By direct substitution we find@2 J#5`.In order to find the central invariant Cˆq we first write the
HO algebra in the conventional form
@aˆ,Nˆ#5aˆ, @aˆ†,Nˆ #52aˆ†, @aˆ,aˆ†#5G
q~Nˆ!, where Gq~Nˆ!5 Q~11Q! ~Nˆ2Q!~Nˆ212Q!, Q5q/~12q!. ~18! Cˆq is then formulated as
Cˆq5aˆ†aˆ1aˆaˆ†1F~Nˆ!,
where
F~Nˆ!5 SN
ˆ21TNˆ1U
~Nˆ2Q!~Nˆ212Q! ~19!
with the coefficients S5(cq12 Q), T5@2cq(112 Q)
22 Q2], and U5(c
q21)(Q1Q2). Here cq is an
undeter-mined eigenvalue of Cˆq. It is easy to see also from Eqs.~18!
and ~19! that there are lower (n50) and upper (n52 J) bounds in the spectrum such that
aˆ†aˆu0
&
J5aˆaˆ†u2J&
J50and
F~2J!2Gq~2J!5F~0!1Gq~0!. ~20!
IV. THE HOMOGRAPHIC q-BOSON REALIZATION OF su„2… DEFORMATION
In the q-boson realization of su~2! deformation, the fun-damental spinor realization is mapped onto a pair of com-muting homographic oscillators as
Iˆ15aˆ1†aˆ2, Iˆ25aˆ1aˆ2 †
, Iˆz5
1
2~Nˆ12Nˆ2!, ~21! where independent algebras for aˆ1 and aˆ2 are given by the analogs of~5!. Using ~5! and ~6! the operators in ~21! can be found to satisfy
@Iˆ6,Iˆz#57Iˆ6, @Iˆ1,Iˆ2#5 fIˆ~Iˆz!2 fIˆ~Iˆz21!, ~22!
where fIˆ~Iˆz!5Q1Q2 Iˆ2Iˆz Iˆ2Iˆz212Q2 Iˆ1Iˆz11 Iˆ1Iˆz2Q1. ~23!
Here Qi5qi/(12qi) (i51,2) and qi’s are the deformation parameters. It is also easy to see that Iˆ251/2(Iˆ
1Iˆ2
1Iˆ2Iˆ1)1Iˆz2 is an invariant of the algebra with eigenvalue
i(i11) where i51/2(n11n2) and iz51/2(n12n2). Hence
As q1 and q2 independently approach unity in the zero
de-formation limit, the deformed algebra in ~22! approaches su~2!.
Now, our aim is to find the invertible nonlinear transfor-mation between the generators Iˆ6,Iˆz and the generators
Jˆ6,Jˆz of the limiting su~2!. We seek an invertible operator
function G (Jˆz) such that@11#
Iˆ15G ~Jˆz!Jˆ1, Iˆ25Iˆ1
†
, Iˆz5Jˆz1const, ~24!
where the constant only depends on the central invariant. The type of deformation in ~22! is not suitable for any Laplace~or Fourier! representation @11# in terms of the pow-ers of Jˆz. However, one can still find the central element Cˆ of this algebra such that
Iˆ1Iˆ25G2~Jˆz!Jˆ1Jˆ25Cˆ 2F ~Jˆz!,
Iˆ2Iˆ15Jˆ2G2~Jˆz!Jˆ15Cˆ 2F ~Jˆz11!,
~25!
where F (Jˆz) is to be found. Since Iˆ is the group invariant, Cˆ is a function of Iˆ only. Therefore its eigenvalue cQ1Q2
only depends on Iˆ and the deformation parameters q1,q2.
Furthermore, the existence of the lowermost and uppermost states on which the action of Iˆ2 and Iˆ1yields zero, respec-tively, implies that Cˆ5F (2J)5F (J11). Using the con-dition@Iˆ6,Cˆ#50, the operator function F (Jˆz) can be easily
found as
F~Jˆz!52 fIˆ~Jˆz21!. ~26!
From ~24!, ~25!, and the su~2! relation Jˆ1Jˆ2 5 1 2(Jˆ 22Jˆ z 21Jˆ z) we finally obtain G2~Jˆz!52 fIˆ~Jˆz21!1Cˆ ~Jˆ22Jˆ z 21Jˆ z! . ~27!
Equations~23!, ~26!, and ~27! define the invertible deforma-tion G (Jˆz).
V. COPRODUCT
One implication of Sec. IV is that the existence of the simple su~2! coproduct
D~Jˆ6!5Jˆ6^I1I^Jˆ6,
D~Jˆz!5Jˆz^I1I^Jˆz, ~28!
and the invertibility of G (Jˆz) guarantee a coproduct law@11#
for Iˆ6,Iˆz. The deforming map defined in Eqs.~24! and ~27!
is, however, not in the class of generalized prototype su(2)q deformations of Curtright and Zachos @11,12#. The
nonpolynomial forms of Gq(Nˆ ) and fIˆ(Iˆz) do not permit a
closed analytic form for the coproduct D(Iˆ6) and D(Iˆz).
However, one can get some hint from the interesting sym-metry displayed by Eq.~22! in the limits of very large ~i.e.,
qi→2`) and very small ~i.e., qi→1) deformations as
limqi→1@Iˆ1,Iˆ2#52 Iˆz,
limqi→2`@Iˆ1,Iˆ2#54eIˆz/Iˆ,
where
Qi52~11e!, e!1. ~29!
Hence in both limits the deformed algebra ~22! behaves like a pure su~2!. In principle, D(Iˆ6) andD(Iˆz) can be obtained
by the application of the deforming invertible transformation found in Eqs.~24! and ~27! ~e.g., see Ref. @11#! as
D~Gˆ!5T@I^T21~Gˆ!1T21~Gˆ!^I#, ~30!
where gˆ5(Jˆ6,Jˆz), Gˆ 5(Iˆ6,Iˆz), and Gˆ 5T(gˆ) is just a
com-pact notation for the transformation in Eq. ~24!. Equation
~30! implies
@D~Iˆ6!,D~Iˆz!#57D~Iˆ6!, @D~Iˆ1!,D~Iˆ2!#5D~Iˆz!.
~31!
Here we notice that D(Gˆ) should behave like D(gˆ) in the symmetric limits ~i.e., qi→1,qi→2`). A possibly existing
simple analytic form of D(Gˆ) might be connected to the closed form~30! by a unitary transformation @12#. However, no explicit and general form for such a transformation is known at the moment.
VI. DISCUSSION
The isomorphism between the homographic oscillator al-gebra ~5! and su~2! has subtle implications in the angular momentum addition theorem and the coproduct law for
Iˆ6,Iˆz. From ~24! and ~25! it is easy to see that
@Iˆ,Jˆ6#5@Iˆ,Jˆz#50 and @Iˆ,Jˆ2#50. These commutations
fur-ther imply an isomorphism betweenuiiz
&
andu j jz&
. In other words, the two basis vectors are parallel to each other al-though they are raised and lowered in different scales on thez axis@19#. We also note that Iˆ is a function of Jˆ only. Let us
now define i and iz as quantum numbers corresponding to a basis set uiiz
&
on which the generators in~22! apply. Then,using ~27! and acting Eq. ~24! on this basis, one obtains
j~ j11!5cQ1Q21Q1Q2
i~i11!
~i2Q1!~i2Q221!, ~32!
which is the desired relation between Iˆ and Jˆ. Here cQ
1Q2 is
the eigenvalue of Cˆ . In the undeformed limit ~i.e.,
Q1,Q2→`) the equivalence of the two algebras requires
cQ1Q2→0.
The arguments presented above guarantee the existence of an implicit coproduct, making it possible to consider Iˆ6and
Iˆz as elements of a quantum deformation of su~2!. The
de-formation parameter is shown to be determined by the total
angular momentum J. Our work is under progress to extend
the arguments presented above to the most general case of quartic oscillator algebra. To the knowledge of the authors, such generalized cases have also been examined recently by Smith @20# as applied to the more general nonunitary case
The stereographic projection is intimately connected with the polar construction of su~2! generators, the quantum phase problem, and, in particular, certain geometrical realizations
@21# of quantum Stokes parameters of polarization @14,21,22#. Nevertheless, the nonunitarity of the azimuthal
phase Eˆfand/or the unavoidable nonzero commutations be-tween the azimuthal and polar phase operators@e.g., see Eq.
~4! here# plague the polar operator construction of su~2!. As
briefly mentioned in Sec. II, the resolution was given by Ellinas@6# by adding a cyclic property to the matrix elements of Eˆf along the Jz axis with a periodicity of (2J11). This
new term does not affect the original spectrum and further makes Eˆf manifestly unitary.
The homographic oscillator representation is physically relevant for its application, particularly in the weak intensity regime of quantum Stokes parameters@14,21#. The quantum phase problem has been studied in the context of an opera-tional point of view by Noh, Fouge´res, and Mandel using two coherent laser sources @23#. More recently this formal-ism has been applied to the case of polarization measurement
of weak fields @14#. Using Poincare´’s stereographic projec-tion, the angular parameters of the polarization ellipse can be mapped conveniently on the Stokes parameters. This has a certain advantage from the operational point of view. The direct measurement of the quantum Stokes parameters might be more relevant in determining the orientation of the polar-ization ellipse both for experimental perspective and the suit-able group properties that quantum Stokes parameters pos-sess. This is where the authors believe that the homographic oscillator introduced here can be linked with the quantum Stokes parameters and polarization measurement. Another dimension of our work in progress is to exploit this physical application.
ACKNOWLEDGMENTS
T. H. appreciates critical comments by Professor C. K. Zachos and Professor S. Stepanov. He is also grateful to Professor O. Viskov for remarks on Sec. III and for bringing Ref. @20# to his attention.
@1# L.C. Biedenharn, J. Phys. A 22, L873 ~1989!; A.J. Macfarlane, ibid. 22, 4581~1989!.
@2# M. Baker, D.D. Coon, and S. Yu, Phys. Rev. D 5, 1429 ~1972!; M. Arik, D.D. Coon, and Yun-ming Lam, J. Math. Phys. 16, 1776~1975!; M. Arik and D.D. Coon, ibid. 17, 524 ~1976!. @3# Ma Zhang-Qi, Yang-Baxter Equation and Quantum
Envelop-ing Algebras~World Scientific, Singapore, 1993!; R.J. Baxter, Exactly Solvable Models in Statistical Mechanics~Academic, New York, 1982!.
@4# A.B. Zamolodchikov and Al. B. Zamolodchikov, Ann. Phys. ~N.Y.! B 120, 253 ~1979!.
@5# M. Chaichian, D. Ellinas, and P. Kulish, Phys. Rev. Lett. 65, 980~1990!.
@6# Demosthenes Ellinas, Phys. Rev. A 45, 3358 ~1992!; J. Math. Phys. 32, 135~1991!.
@7# R.M. Mir-Kasimov, J. Phys. A 24, 4283 ~1991!; M. Arık and M. Mungan, Phys. Lett. B 282, 101~1992!.
@8# D. Bonatsos, C. Daskaloyannis, and P. Kolokotronis, J. Phys. A 26, L871~1993!.
@9# Ian L. Cooper and Raj K. Gupta, Phys. Rev. A 52, 941 ~1995!. @10# O¨. F. Dayı and I. H. Duru, J. Phys. A 28, 2395 ~1995!. @11# T.L. Curtright, Proceedings of the Argonne Workshop on
Quantum Groups, edited by T. Curtright, D. Fairlie, and C. Zachos ~World Scientific, Singapore, 1991!, p. 72; C.K. Za-chos, in ibid., p. 62.
@12# T.L. Curtright and C.K. Zachos, Phys. Lett. A 243, 237 ~1990!; T.L. Curtright, G.I. Gandour, and C.K. Zachos, J. Math. Phys. 32, 676~1990!.
@13# M.M. Nieto, Phys. Rev. 167, 416 ~1963!; J.M. Jauch and F.
Rohrlich, The Theory of Photons and Electrons ~Addison-Wesley, Reading, MA, 1955!.
@14# T. Hakiog˘lu, A.S. Shumovsky, and O. Aytu¨r, Phys. Lett. A 194, 304 ~1994!; L.L. Sanchez-Soto and A. Luis, Opt. Com-mun. 105, 84~1994!.
@15# A. Vourdas, Phys. Rev. A 41, 1653 ~1990!. @16# M. Arık ~unpublished!.
@17# E. I˙no¨nu¨ and E.P. Wigner, Proc. Natl. Acad. Sci. ~USA! 39, 510 ~1953! @a similar approach to the classical contraction limit was also studied in Y.J. Ng, J. Phys. A 23, 1023~1994!#. @18# M. Arık et al., Z. Phys. C 55, 89 ~1992!.
@19# In the case of su~2!qdeformation, there is no known invertible
analytic deforming map to obtain the deformed case from the su~2! generators. Therefore the obvious interconnectedness be-tween the parallel vectors uiiz& and u j jz& is not trivial in the
case of su~2! → su~2!q.
@20# S.R. Smith, Trans. Am. Mater. Soc. 322, 285 ~1990!. @21# Poincare´’s map from the Stokes parameters on S2 to the
ex-tended complex planeC is nothing but the conventional stereo-graphic projection applied on the Stokes vector SW5(S1,S2,S3) for fully polarized light~i.e., uSWu51). For the original reference see, H. Poincare´, The´orie Mathe´matique Lumie´re 2, 275~1892!; also J.M. Jauch and F. Rohrlich, Ref. @13#, Chap. 2.
@22# R.M. Azzam, Ellipsometry and Polarized Light ~Elsevier, Am-sterdam, 1989!.
@23# J.W. Noh, A. Fouge´res, and L. Mandel, Phys. Rev. Lett. 67, 1426~1991!; Phys. Rev. A 45, 424 ~1992!; 46, 2840 ~1992!.